Climate-assisted persistence of tropical fish vagrants in temperate marine ecosystems

Climate change is reshaping the function and composition of ecosystems globally. Marine species often respond by shifting distributions poleward or into deeper, cooler waters, but extreme events such as marine heatwaves can disrupt the pace and trajectories of these responses. Heatwaves have caused major ecological impacts (e.g., kelp die-offs, coral bleaching, mass mortalities), yet they can also propel tropical and subtropical organisms into temperate systems, contributing to the tropicalization of temperate marine environments. Despite many observations of poleward expatriation, the ecological and evolutionary mechanisms enabling persistence of tropical vagrants in temperate environments remain poorly understood. Whether species remain after heatwaves likely depends on thermal sensitivities of key traits, availability of resources, competition, and potential for rapid evolution. Given increasing frequency of extreme events and rapid niche shifts in introduced species, it is crucial to determine whether tropical marine vagrants will establish permanently under future warming. Western Australia’s strong latitudinal SST gradient and the 2010/2011 extreme marine heatwave (driven by a strengthened Leeuwin Current) provided a natural experiment: tropical Black Rabbitfish (Siganus fuscescens) were introduced into temperate habitats and reproductive adults were observed near the previous range margin. This study evaluated: (1) population connectivity and genetic differentiation among tropical residents and temperate vagrants to infer heatwave-enhanced migration and panmixia; (2) dietary flexibility via DNA metabarcoding of stomach contents to assess adaptation to novel resources; and (3) climate-based isotherm projections to estimate potential overwintering, spawning, and further range expansion by 2100. The overarching hypothesis is that favourable future thermal conditions, coupled with connectivity and dietary versatility, could convert vagrancy into permanency with consequences for native temperate communities.

Population genomics: Fin clips or gill tissue were collected from 223 Siganus fuscescens individuals across Western Australia from 2013–2017 (Kimberley n=40, Pilbara n=36, Exmouth Gulf n=9, Coral Bay n=7, Shark Bay n=40, Cockburn Sound n=51, Wanneroo Reef n=40). After quality control, 220 individuals remained. DNA extractions used Qiagen DNeasy Blood & Tissue Kit with modified buffers and partial automation (QIAcube). SNP genotyping followed DArTseq (reduced-representation with PstI–SphI and PstI–NspI digestion; Illumina HiSeq2500). Initial raw data yielded ~168,000 SNPs (24.71% missing). Filtering retained loci genotyped in ≥95% of individuals; coverage 20×–200×; minor allele frequency ≥0.05; heterozygosity ≤0.75. After removing monomorphic and low-coverage loci, 6,505 SNPs across 220 fish remained. Loci out of HWE and LD (after Bonferroni) were removed (−826), leaving 5,679 SNPs. Outlier detection (Outflank; 5% left/right trim; min heterozygosity 0.001; FDR 5%) identified 172 outliers and 5,507 putatively neutral loci. Genetic diversity (allelic richness, expected/observed heterozygosity) was computed with diveRsity (10,000 permutations). Pairwise FST (neutral SNPs) with StAMPP (10,000 permutations; BY-FDR correction); GST with diveRsity (1,000 iterations). Population structure was inferred with fastSTRUCTURE (K=2–12; internal model selection), and DAPC (adegenet) used for visualization. Isolation-by-distance was tested via Mantel tests between linearized FST and oceanic geographic distances at 0, 1, and 10 m depths (vegan; 100,000 permutations; distances from marmap). Relative gene flow and directionality among sites were estimated using divMigrate (based on GST; 10,000 bootstraps) and visualized via heatmaps (corrplot). Dietary DNA metabarcoding: Stomach contents from a subset of 34 individuals were analyzed to characterize phytoplankton and algal diet: tropical/subtropical residents (Coral Bay n=7, Shark Bay n=10) and temperate vagrants (Wanneroo Reef n=8, Cockburn Sound n=9). Some individuals were excluded due to low sequencing or amplification failures (final tropical sample sizes reduced by CB −3, SB −4). Stomachs were processed within 12 h; DNA extraction used Qiagen Mini Stool Kit with two bead-beating steps. Targeted the plastid 23S rRNA gene using modified universal UPA primers (p23SrVf1/p23SrVr1; 370–400 bp amplicons). qPCR optimization determined dilution and annealing temperature; final qPCRs performed in duplicate with 8-bp MID-tagged primers; libraries pooled equimolarly, adapter-ligated (NEBNext Ultra), size-selected (250–600 bp; Pippin Prep), and sequenced on Illumina MiSeq (V2, 2×250). Bioinformatics: Paired reads merged (MiSeq Reporter), demultiplexed (Geneious), quality-filtered (adapter/primer exact matches), dereplicated (USEARCH v10), filtered for error rate ≤1% and length ≥100 bp, abundance-filtered (≥2 reads), chimera-checked, clustered into OTUs at 97% similarity (USEARCH), normalized to 30,000 reads/sample. Post-clustering curation with LULU reduced 1,337 initial OTUs to 735; 17 OTUs detected in controls (≥2 reads) removed (after BLAST verification), yielding 718 OTUs. BLASTn (E≤0.001; 100% length; ≥97% identity; best-hit edge 5%; overhang 25%; bit score >620) against NCBI assigned taxonomy; non-bacterial/eukaryotic or unassignable sequences removed. Final table included 86 OTUs; those with <10 reads were discarded, yielding 78 OTUs for analyses (cyanobacteria, dinoflagellates, diatoms, microalgae, and green/red/brown macroalgae). Visualization used Circlize (relative abundance) and heatmaps (phyloseq; Bray–Curtis; top 30% abundant OTUs). Community analyses: nMDS (Bray–Curtis; stress=0.12), PERMANOVA (100,000 permutations), PERMDISP2 (100,000 permutations), and indicator species analysis (IndVal; 100,000 permutations; BH correction) identified region-specific dietary items. Because Ecklonia radiata 23S sequence was absent in NCBI, regional E. radiata tissue was sequenced to generate a voucher (GenBank MW752516) and used to confirm kelp reads in stomach contents. Climate observations and projections: Historical SST climatologies for 1900–1909 and 2008–2017 from HadISST (monthly, 1°) and daily 5-km NOAA Coral Reef Watch CoralTemp v1.0 for 2011 heatwave anomalies. Near-surface (0–25 m) ocean current anomalies for austral summer 2011 from SODA v3.3. CMIP5 model outputs (ensemble of 11 models; r1i1p1) were used for historical GHG, RCP4.5, and RCP8.5 scenarios; monthly SST anomalies computed relative to 2008–2017 and added to HadISST climatology for mean annual and minimum monthly mean (MiMM) fields. Ensemble means were used to plot decadal averages and ranges of the 20 °C mean-annual and 17 °C MiMM isotherms for past (1900s) and future (2090s) periods. Analyses performed in MATLAB 2012b.

- Genomics: Across 5,507 neutral SNPs, there was limited to no significant genetic differentiation among most tropical and temperate sites (pairwise FST and GST largely non-significant; Supplementary Tables S1–S2). fastSTRUCTURE supported K=1 for neutral loci; K=2 for outlier loci, likely reflecting sampling bias (low Exmouth Gulf/Coral Bay sample sizes) or potential adaptive divergence not linked to specific genes. Migration analyses showed relatively high, non-directional gene flow between tropical residents and temperate vagrants, indicating panmixia facilitated by the Leeuwin Current; Exmouth Gulf and Coral Bay showed some isolation (reduced exchange). No isolation-by-distance pattern at local scales (Mantel tests; Supplementary Fig. S1). Genetic diversity (heterozygosity) was similar between vagrants and tropical residents, implying sufficient standing variation for adaptation in expansion zones. - Diet: DNA metabarcoding identified 78 phytoplankton and algal OTUs in stomach contents, with the top 30% of reads dominated by red and brown macroalgae. Diet included taxa spanning tropical-temperate distributions (e.g., Asparagopsis taxiformis, Champia parvula, Lobophora variegata, Padina australis, Spyridia filamentosa, Tolypiocladia glomerulata), tropical Coelothrix irregularis, invasive Sargassum natans, and habitat-forming kelp Ecklonia radiata. Epilithic resources (diatoms, dinoflagellates, microalgae, cyanobacteria including Synechococcus) were also consumed. nMDS indicated limited overlap between tropical (Coral Bay, Shark Bay) and temperate (Wanneroo Reef, Cockburn Sound) diets; temperate sites were completely segregated. PERMANOVA: df=2, F=6.56, R^2=0.47, p=1×10^-5; PERMDISP2: df=2, F=2.43, p=0.09, indicating significant compositional differences not due to dispersion. Indicator species analysis found ~18% of resources were unique to a region/combination; red macroalgae most strongly characterized regions; brown algae (Ectocarpales) and kelp E. radiata characterized a temperate region. - Projections: The MiMM 17 °C and mean annual 20 °C isotherms (thresholds for overwintering and spawning) are projected to shift substantially poleward by 2100 under both RCP4.5 and RCP8.5. Under RCP8.5, the isotherms may encompass the entire southern Australian coast by 2100; under RCP4.5, suitability remains largely on the western coast. These conditions suggest S. fuscescens vagrants could survive and self-recruit up to ~2,400 km beyond their post-heatwave southern range limit by 2100, contingent on availability/connectivity of seagrass and macroalgal juvenile habitats. The cohort of vagrants included maturing and reproductively mature individuals, supporting potential self-recruitment.

The study demonstrates that three mechanisms jointly facilitate the persistence and potential establishment of tropical Siganus fuscescens in temperate Western Australian ecosystems: (1) strong, large-scale genetic connectivity with minimal population subdivision between tropical sources and temperate vagrants, likely enhanced by the Leeuwin Current and the 2010/2011 heatwave; (2) dietary versatility spanning diverse phytoplankton and macroalgal resources, including regionally important kelps (Ecklonia radiata), enabling exploitation of temperate food webs; and (3) projected warming that expands thermal niches (overwintering and spawning isotherms) poleward, creating climatically suitable habitats by century’s end. The lack of significant genetic structure and high gene flow imply continued replenishment of temperate vagrant populations and sufficient standing genetic variation to support adaptation in new environments. Distinct regional diets indicate plasticity and local resource use rather than strict specialization, likely supported by a flexible gut microbiome that facilitates algal fermentation and digestion. Climate projections indicate that without substantial emission reductions, southern Australian coasts will become suitable for survival and reproduction, potentially converting episodic vagrancy into self-sustaining populations. Ecologically, increased herbivory pressure from established rabbitfish could compromise kelp forests of the Great Southern Reef, with cascading effects on biodiversity, fisheries, and ecosystem services. Establishment success will depend on the distribution and resilience of recruitment habitats (seagrass, macroalgae), the thermal tolerance of native macrophytes, and the ability of rabbitfish to utilize alternative nursery habitats during resource fluctuations.

This work integrates genomics, diet metabarcoding, and climate projections to show that tropical Black Rabbitfish can persist and potentially establish in temperate Australian waters under ongoing warming. Key contributions include evidence of panmictic connectivity between tropical and temperate populations, demonstration of broad dietary breadth including temperate kelp resources, and projections that future thermal regimes will support overwintering and spawning along increasingly poleward coastlines. These findings imply heightened risk to kelp-dominated ecosystems on the Great Southern Reef through intensified herbivory. Future research should: (1) conduct temporal genomic monitoring to detect adaptive responses and changes in connectivity; (2) expand dietary and stable isotope analyses across seasons and life stages to quantify resource use and trophic impacts; (3) investigate gut microbiome function in digestion across temperature regimes; (4) couple biophysical larval dispersal models with habitat suitability and landscape connectivity for seagrass/macroalgae; and (5) assess community- and ecosystem-level consequences of increasing rabbitfish herbivory under different climate scenarios.

- Tropical diet sample sizes were small (4–7 individuals per tropical site after exclusions), limiting power to assess local dietary adaptation and potentially biasing community comparisons. - The outlier SNP K=2 signal may reflect sampling bias (low n at some tropical sites), and specific adaptive loci were not identified (no sequence alignment for the top SNP candidate). - Dietary metabarcoding targeted phytoplankton and algae and may underestimate non-target diet components; index hopping and rare OTUs were minimized but cannot be excluded entirely. - Climate projections used CMIP5 ensemble means and isotherm thresholds as proxies for overwintering and spawning; actual establishment depends on additional biotic and abiotic factors (e.g., habitat connectivity, prey availability, species interactions) not fully modeled. - Inference of self-recruitment is based on temperature suitability and maturity status rather than direct larval/juvenile recruitment observations in all projected regions.